Full exam for Data Science and Security for Mobility in the Mobility Engineering degree programme at Politecnico di Milano. The document covers: Example Exam Questions January 13, 2020 Student ID: Name: Answer the questions in the space provided. If you run out of room, you can use the spare pages at the end, (but remember to clearly mark which question you are answering). 1. What does the parameter k in k-Means control?
Full exam for Data Science and Security for Mobility in the Mobility Engineering degree programme at Politecnico di Milano. The document covers: Example Exam Questions January 13, 2020 Student ID: Name: Answer the questions in the space provided. If you run out of room, you can use the spare pages at the end, (but remember to clearly mark which question you are answering). 1. What does the parameter k in k-Means control?
Import quality: text was extracted directly from the original document.
Representative passages recognised in different parts of the material. The full extracted text remains available to search, while this compact preview makes the page easier to read.
Example Exam Questions January 13, 2020 Student ID: Name: Answer the questions in the space provided. If you run out of room, you can use the spare pages at the end, (but remember to clearly mark which question you are answering). 1. What does the parameter k in k-Means control? 2. What happens to the within-cluser dispersion (Sum of Squares) as the parameter k increased? 3. How can one choose an appropriate value for k? 4. Compare the memory requirements of different clustering algorithms. Hint: you can (but don’t need to) use the complexity notation: O() 1 5. Continuing the discussion above, which clustering algorithm might struggle to cluster a dataset contain- ing 1 million items? Why? 6. What is a k-d tree and what are they used for? 7. Is a Decision tree considered a linear classifier ? Why or why not? 8. Which classification algorithm directly maximises the buffer region between instances of the two classes? 9. Will increasing the maximum depth of a decision tree make the classifier more or less stable (i.e. decrease or increase its variance)? Page 2 10. List some approaches for dealing with missing values. 11. What is the difference between supervised and unsupervised learning? 12. Is anomaly detection a form of supervised or unsupervised learning approach? 13. Principle Component Analysis (PCA) can be used to visualise high dimensional data by plotting the first two (or three) principle components. What is the first principle component? 14. What should be done before running PCA? 15. Bonus question : Is Mark Carman the best lecturer you’ve ever had? Explain (carefully ;-) why or why not. Page 3
First page of the document.
